Defining data science, AI and machine learning in financial services

  • Discuss how AI and machine learning can be applied to financial services.
  • Understand its benefits and limitations within the advisory field.
  • Explain the benefits and drawbacks of this technology for the advice process.

Building up a data set of evidence from history, one can create a predictive model based on hidden, complex underlying structural relationships buried within historical data sets.

Elsewhere, particular investment strategies might use trading algorithms to set rules around buying-and-selling behaviours, based on share price movements, for example. We are witnessing the rise of quantitative strategies that rely on such rules-based decision-making and I expect this to develop further, especially as downward pressure on costs shows no sign of abating.

Other AI applications

Other applications might use AI to develop a creative process of understanding human beings. If you can use AI to determine clients' needs, goals, understand their circumstances and apply it to their environment to develop a plan for their financial future, that might be venturing towards strong AI.

I believe to truly achieve a bespoke financial plan is an incredibly sophisticated process that involves huge amounts of human intuition, specific knowledge and empathy. As such, I cannot see that being outsourced completely from a human to a computer any time soon.

There are aspects of this process, however, that can do some of the early-stage filtering, or client segmenting activity, which could then be picked up by a human and apply their more nuanced thinking to the individual’s case, which may involve overriding the automated path first taken.

We are seeing this happening to varying degrees of success at the moment, with so-called robo-advice models. This might include automating part of the advice process to make certain suggestions or recommendations to act, or not act, on a particular product if the client met certain criteria. The issue with many of these models is that I see them as a very blunt tool that may not give the best client outcomes in all circumstances.

While one of the major potential upsides to the increased use of technology forming or supporting at least part of the advice process is that in bringing costs down, theoretically you can democratise advice to an extent and expand its availability to a wider audience. And the more we can automate and optimise that process, the quicker and slicker that type of operation will be rolled out across the market. 

However, my view is that there needs to be a narrow band of applicability for those automated processes, so it is important to keep these aspects fairly simple, transparent and easy to understand. These could then be supported and built on by more sophisticated requirements with some human intervention. What I do not believe is that we ought to rush to build very complex robotic advice processes to replicate large portions of the advice journey because we will very quickly end up with things we do not understand.